As processing speeds continue to improve and data storage becomes ever less expensive, many sophisticated applications that were previously only available on mainframe or desktop computers have been ported to laptop computers and other portable electronic equipment. Many applications have even been ported to hand held electronic devices as well, including hand held computers, digital telephones, personal digital assistants (PDAs), and so forth. For example, personal databases with limited search capabilities are now included in cellular phones, and word processing can now be performed in PDAs.
There are, however, several applications that are presently difficult or impossible to realize on hand-held electronic devices, and are only poorly realized even on larger systems such as desktop computers. Due to the large volumes of data involved, and the need to process at very high speeds, a particularly difficult application is voice recognition. Some attempts have been made in that direction, but all of them suffer from one or more disadvantages.
At the low end, limited word or phrase recognition capabilities are sometimes provided in cell phones. Such systems can usually recognize only a few words (e.g., the numerals 0-9, and specialized key words such as a person's name, or the commands “dial” or “open file patentapp.doc”). Such systems are particularly advantageous where only rudimentary recognition capabilities are needed, or where only very limited data storage capability or computing power is available. However, an obvious shortcoming of the word or phrase recognition systems is that the usability is limited to a small, preprogrammed vocabulary, and at most a few custom words. Moreover, word or phrase recognition systems often fail to recognize personal speech pattern or accents.
At the higher end, speech recognition programs are currently available for operation on laptop computers. As used herein both “speech recognition” and “word or phrase recognition” are considered to be categories of voice recognition. “Speech recognition”, however, is limited to systems having a vocabulary of at least 200 words, and where individual words are interpreted in the context of surrounding words. For example, speech recognition would correctly interpret phrases such as “I have been to the beach” whereas a word or phrase recognition system may substitute “bean” for “been”.
As with other computer software application, most of the development effort is being directed towards porting the more sophisticated speech recognition to smaller and smaller devices. It may well be that within a decade the goal of true speech recognition will be available on even hand-held electronic devices.
What is not presently appreciated, however, is that porting of sophisticated software to portable electronic devices may not be desirable. Cell phones, for example, need only relatively rudimentary electronics to support the required communications, and placing sophisticated storage and processing in cell phones may be a waste of money. Moreover, no matter how sophisticated the software and hardware becomes in hand held and other portable devices, there will always be a perceived need for additional capabilities. Larger or specialized vocabularies may be desired, as well as recognition capabilities for different accents and languages, and perhaps even language translation capabilities. Still further, it is impractical to install voice recognition in all the myriad types of devices that may advantageously utilize voice recognition. For example, voice recognition may be useful in VCR and CD players, kitchen and other household appliances such as toasters and washing machines, automobiles and so forth.
Thus, while it has been known to translate information in a first sense modality and language into a second sense modality and language on a single local computer, it has not been appreciated to perform the translation in a “remote computing” manner, thereby concentrating the computing power in a cost effective manner. Consequently, there is a need to provide voice recognition capabilities, and especially speech recognition capabilities, to myriad electronic devices without actually installing all of the required hardware and software in all such devices.